On Tue, Jun 12, 2012 at 3:17 PM, Alan Bell<alanbell@xxxxxxxxxx> wrote:
for reference, it would look like this
http://people.ubuntu.com/~alanbell/screenshots/unityretina.png
(scaled down by a factor of 4)
I see three routes:
1) Two or more sets of pixel-based sizes for designs and toolkits,
with ranges of DPI for application.
2) Point based designs.
3) Millimeter based designs.
#1 is easiest. It's always "exact" from a toolkit point of view.
Design on graph paper is no problem. It probably covers most cases to
have two sets: the current and another with doubled lengths. I think I
read that, as a matter of practice, this is what Apple's point-based
guidelines do since they control the hardware.
#2 is what Apple has explicitly done. #3 is the same as #2, but using
SI units. The tricky thing is keeping things pixel perfect at low-res
when you don't control the hardware. ("Retina" claims aside, I believe
it's still tricky at high-res, but not as tricky.) Ideally the choice
is pushed into the toolkit and code developers can be given toolkit
units, whatever they may be. The toolkit should be able to smooth out
the differences between displays. Basically I'm saying claim #2 or #3
but actually do #1.
Between #2 and #3, I prefer SI units.
Maybe there's a #4: Microsoft's Metro grid. Looking for what they use,
I found an article I read about this:
http://bjango.com/articles/everythingisagrid/ . Therein it is
indicated that Apple claims #2 and does #1.
Whatever the case, I don't know that there's a tool to make good
on-screen design easy. I know about Glade and some wireframing apps,
but I still see designers drawing by hand, so I take it the tools
aren't good enough yet. (Or maybe people are picking the wireframing
apps that look hand-drawn?) Be it RAD or wireframing tools, they all
seem to focus too much on laying out controls in dialogs and not
enough on designing the core information manipulation parts of
applications. It's not surprising we don't see more direct
manipulation when the tools don't help make it so.
